A full-day workshop on Humanoid Technologies
With a few notable exceptions, experimental robotics was very often based on industrial robots (the good old PUMA family for instance), modified to be interfaced to standard computers and completed by commercially available sensors integrated through a plethora of different interfaces. The controller was often centralized. The approach has changed now: it is common to find teams developing robots where the mechanics, the controller and the sensors are integrated into a complete mechatronic design. The problem at large is now how to make this technologies effective and in finding ways to collate the effort of the single into a coherent and harmonious technology that can be shared and reused, transferred, and improved at a faster pace.
Theme 1, morning, State-of-the-art approaches
Reliability is now an issue; experiments last longer than the typical 10-second benchmark demonstration which was customary in the past. In this new scenario, the old problem of interfacing and integrating different subsystems becomes even more significant. From the sensing and control point of view the paradigm of a centralized computer is not sufficient and, in fact, the emerging approach is based on the development of distributed sensors and control networks. System design, in this context, the development, maintenance, and upgrading of such robotic platforms require tools and perhaps new methodologies. We would like to foster a debate on aspects like (but not limited to):
- How to exploit distributed and parallel structures for sensing and control;
- How to develop new algorithms and protocols;
- How to integrate heterogeneous components flexibly and with an eye at scalability;
- Whether there is a need for specialized hardware and software (networks and protocols);
- Whether we require programmable vs. hardware devices, industrial state-of-the-art vs. custom chips;
- Whether there is any technological breakthrough along the way: e.g. self-reconfiguring connections.
Theme 2, afternoon, Motivations for Humanoid Technologies
A Humanoid Consortium: Ways to advance humanoid technologies. The realisation of a humanoid robot hinge upon a great deal of technical "know-how", impact on progress highly depends on such "know-how". The idea of this afternoon theme is to draw together research groups in the area of humanoid robotics in examining the past and recent advancements in humanoid technologies, and to propose ways to fast track towards a better future for the development of "humanoid technologies" in advancing at large the research field of Humanoid Robotics. This second theme draws on the morning session by analysing and proposing ways of effecting this large-scale speed up and integration: having clarified the requirements, we would like to discuss how to move forward, topics are (but not limited to):
- Licensing schemes, the contribution of industry, funding agencies;
- Sharing of components, recycling, software and hardware development of recyclable modules and standards;
- Standardization (e.g. IEEE standard?);
- Education and international links between projects, cooperation.